Modelling of crop yields and N2O emissions from silty arable soils with differing tillage in two long-term experiments

The choice of tillage system affects crop growth and soil nitrogen dynamics. Models help us to better understand these systems and the interaction of the processes involved. Objectives were to test a calibration and validation scheme for applications of the denitrification–decomposition (DNDC) model to describe a long-term field experiment with conventional tillage (CT) and reduced tillage (RT) at two sites (G and H, silty Haplic Luvisols) near Göttingen, Germany (G-CT, G-RT, H-CT, H-RT). Crop growth of field bean (Vicia faba L.) and winter wheat (Triticum aestivum L.) as well as soil water dynamics and nitrous oxide (N2O) emissions were determined for two subsequent years. A model test was performed based on a model parameterization to best describe the case G-CT. This parameterization was then applied to the other cases as a retrospective simulation. Results of model variant v1 (no parameter optimization) indicated that soil water contents were not accurately simulated using the DNDC default values for a silt loam. After successful calibration of the soil water flow model using modified water-filled pore spaces at field capacity and wilting point and a modified hydro-conductivity that led to a good fit of the measured water content data, grain yields were markedly underestimated and modelled N2O emissions were too large (v2). An optimization of the crop properties (maximum grain yield, N fixation index, thermal degree days, transpiration coefficient) was essential for a better match of measured yields (v3). Further adjustments in the model (v4) were required to better match cumulative N2O emissions: reducing the initial soil organic carbon content and mineralization rates. Predictions of crop yields and annual cumulative N2O emissions using model variant v4 were fairly accurate for the reduced tillage system G-RT and also for the second field experiment H-CT and H-RT, but annual distributions of N2O emissions were not. Overall our results indicate that site specific calibration was an essential requirement for the silty German sites, and that the pedotransfer functions and denitrification sub-model of DNDC may need further improvement.

Research highlights
► The DNDC model is a very user-friendly model, requiring inputs that are generally easy to obtain.
► Our results indicate that site specific calibration was an essential requirement for the silty German sites.
► The pedotransfer functions and denitrification sub-model of DNDC may need further improvement.